Hi Robert, Thank you for bringing the discussion to the mailing lists.
#2 seems like a good option, if we can reach consensus with the Bahir community. However, should we also be considering “staging” (perhaps under “flink-contrib”?) a connector contribution until more committers can maintain it, and then move it into Flink? I feel like that we shouldn’t be redirecting all connector contributions away from Flink simply based on committer capacity. Some new connectors may have a wide user pool and can be staged first. So, I feel like redirection of connector contributions should go in 3 ways: 1. Redirect to option #1 or option #2 2. Stage first. We can officially move them to “flink-streaming/batch-connectors” once more committers can maintain it. 3. Directly contribute it to “flink-streaming/batch-connectors” as an officially supported connector, if initially there’s already committers willing to maintain it. As for option #3, from the discussion thread in "Externalizing the Flink connectors”, it seems like that the solution is mainly to provide more maintenance flexibility for our supported connectors, instead of solving the issue of committer capacity for new connectors, correct? Regards, Gordon On August 9, 2016 at 4:43:56 PM, Robert Metzger (rmetz...@apache.org) wrote: #2 Redirect the contribution to Apache Bahir. A recently created Apache project out of Apache Spark, to host some of their streaming connectors.
